(210b) Finding a Needle in an Ocean of Data with Digital Twins: An Oil & Gas Perspective on Analytics

Authors: 
Analyzing industrial data requires a fundamentally different approach than consumer data. This is due to a confluence of several factors. Compared to traditional ERP systems that store transactional data like logistics reports, maintenance logs, inventory logs, etc., Oil & Gas industrial data is 10-100X more voluminous, is collected at 100-1000X more frequently and is 5-10X more diverse. The hybrid modeling framework to build Digital Twins that combines physics-based models with deep learning techniques will be highlighted with specific application to system level optimization. We will present novel application of Digital Twins and deep learning models for anomaly detection in process models, network modeling and process optimization.